Fig. S3.

Open-loop simulation controls. (A) Comparison of experimental results with simulations in which agents are not dependent on any external variables. Shown are representative time courses of experientially observed (Observed) and simulation (Simulated) cell densities over the course of the experimental time window. (B) Average run distance in a 2-cells/μm² sliding window for simulations in which agent behavior is dependent on local cell density (blue lines) and experimental results (red lines). Dashed lines indicated bootstrapped 95% mean confidence intervals. (C) Average (solid lines) and SDs (dashed lines) of the percentage of cells inside aggregates for experimental (red lines) and simulations in which agent’s behavior is driven by time since the beginning of the experiment and local cell density (blue lines) or only local cell density (black lines). (D) Average run duration in a 20-min sliding window for all runs (persistent and nonpersistent) from experimental results inside (black lines) and outside aggregates (blue lines) and open-loop simulations in which agent behaviors were chosen dependent on time since the beginning of the experiment. Green lines indicate agents outside aggregates, and red lines, inside. Dashed lines are as in B. (E) Average run distance in a 10-μm sliding window for simulations in which agent behavior depends on orientation to nearest aggregate. Purple (experimental results) and blue (simulation results) lines indicate runs oriented toward [($\text{cos}\left({\beta }_n\right)>0,$ Fig. S1C] the nearest aggregate centroid. Green (experimental results) and black (simulation results) lines indicate runs pointed away $\left(\text{cos}\left({\beta }_n\right)<0\right)$ from the nearest aggregate centroid. Negative distances indicate that the run began inside the aggregate. Dashed lines indicated 95% bootstrap confidence intervals. (F) Average (solid lines) and SDs (dashed lines) of the percentage of cells inside observed aggregate boundaries for experimental (red lines) and simulations with a biased walk and with (blue lines) or without (green lines) time since the beginning of the experiment as a dependence for choosing the agent’s run state, speed, and duration. (G) Average (solid lines) and SDs (dashed lines) of percentage of cells inside the aggregates in simulations (black lines) in which agents chose their run state, duration, and speed dependent on orientation and distance to the nearest aggregate when the agent was inside an aggregate (C1) or within 25 μm (C2), 50 μm (C3), or 100 μm (C4) of the aggregate boundary. When outside the cutoff distance (C1–C4), no aggregate dependence was used to choose agent behaviors. Blue lines indicate simulations in which aggregate distance and orientation are always included in choosing agent behaviors. Red lines indicate experimental results.